High-Sensitivity MEMS Shear Probe for Autonomous Profiling Observation of Marine Turbulence

Abstract

Autonomous profiling observation of full-depth marine turbulence is very important for ocean research. Anisotropic turbulence near the boundary layer needs to be observed well. However, there is lack of high-spatial-resolution and high-sensitivity methods to fulfill vector turbulence observation. Hence, a highly sensitive bullet-headed MEMS shear probe for observing ocean vector turbulence is developed in this manuscript. The sensing mechanism, design and fabrication are demonstrated in detail. In order to meet the bandwidth requirements for observing ocean turbulence, we perform wet-mode simulations of probe structure to achieve an eigenfrequency of 490 Hz. Through sensitivity calibration experiments, it achieves a sensitivity of 4.84 × 10⁻² V·m·s²/kg, which is much higher than those of shear probes reported previously. In addition, the vector test validates that the probe can measure ocean vector turbulence. The results show that the proposed probe is promising in autonomous profiling observation of marine turbulence.

1. Introduction

Turbulent mixing is the source force of ocean macroscopic phenomena including heat transferring, mass transferring and energy transferring [1]. In terms of physical structure [2], turbulence can be regarded as a flow formed by the interaction of vortices with different scales. The size of these vortices and the direction distribution of the rotation axis are random [3]. Large scale vortices constantly obtain energy from the mainstream. Through the interaction between vortices, the energy is gradually transferred to small vortices and is finally dissipated into fluid heat energy [4]. At the same time, due to the effect of boundary, disturbance and velocity gradient, new vortices are constantly generated, which constitute turbulent motion. Unlike uniform velocity profiling of laminar flow, turbulence has shear flow velocity profiling. The gradient of velocity profiling indicates the intensity, shear stress, energy and dynamics characteristics of ocean turbulence [5]. In order to study the formation and mixing process of turbulence, it is important to observe fine structure evolution of ocean turbulence. Remote sensing methodologies through optical/microwave [6] or scanning Lidar [7] are reported to observe turbulence of the upper sea. Turbulence profiler is the most effective method to observe ocean full-depth vertical turbulence [8]. The structure and working sketch of a free-fall vertical profiler are demonstrated in Figure 1. The profiler can free fall with uniform velocity after being released from the ship. Turbulence data at vertical routine including upper mixed layer, middle layer and seabed boundary layer could be acquired. After ejecting load weight, the profiler could float upward automatically and arrive at the sea surface. The data could be transmitted to an airplane or satellite.

Anisotropic turbulence could be produced by density stratification, especially near the boundary layer, including the air–sea boundary layer, seabed boundary layer, pycnocline, etc. [9,10]. High resolution vector evolution data of turbulence are urgently needed for cognitive improvement in aspects such as wave breaking of the sea surface, double diffusion in the ocean, and deep-sea circulation structures [11]. A development map of the shear probes is demonstrated in Figure 2. A shear probe is the most commonly used instrument for turbulence observation. A platinum hot wire probe was the first sensor used in marine turbulence observation [12]. It can be severely affected by environmental temperature fluctuations and has a narrow bandwidth, making it obsolete in this field. In order to perform marine turbulence observation, airfoil shear probes based on piezoelectric ceramic were developed. N. Oakey from the Bedford Institute of Oceanography proposed a kind of airfoil shear probe [13]. M. Moum from Oregon State University reported a kind of shear probe with a diameter of 3.2 cm and a sensitivity of 0.37 × 10⁻⁴ V·m·s²/kg [14]. H. Prandke et al. from ISW washer company developed a PNS shear probe [15], which had a diameter of 8 mm and a sensitivity of 5.16 × 10⁻⁴ V·m·s²/kg. T. Osbern from Rockland company reported an SPM shear probe with a diameter of 0.95 cm and a sensitivity of 0.57 × 10⁻⁴ V·m·s²/kg [16]. S. Wang et al. from Tianjin University reported an airfoil shear probe with an optimized cantilever beam and piezoelectric ceramics sheet [17]. W. Gao et al. from Northwestern Polytechnical University proposed a sensor for fluid shear stress measurement [18]. However, these airfoil shear probes based on piezoelectric ceramic could only measure unidimensional turbulence. To achieve vector observation, two unidimensional shear probes are usually orthogonally arranged. This will not only bring installation misalignments but will also be unable to achieve single-point multidimensional synchronous observation.

Therefore, a shear probe that can realize vector and high-resolution turbulence observation are needed. In our previous work, several kinds of underwater shear probes were proposed [19,20]. In order to improve the sensitivity, a kind of MEMS shear probe based on bullet-headed cilium and crossbeam is developed. It also has the capacity of vector detection and high-spatial resolution.

2. Materials and Methods

2.1. The Principle of Probe

The MEMS shear probe is inspired by a lateral line of fish. The whole probe has a cilium-crossbeam structure (Figure 3). The cilium is used to receive the turbulence signal and then drive the cross beam to transform. Eight varistors are placed on the cross beam to form two Wheatstone bridges (Figure 4). Based on the piezoresistive effect, the varistor value on the beam changes, and the Wheatstone bridge brings voltage output. Two Wheatstone bridges are arranged vertically, which enables the probe to detect turbulence signals in two different directions.

The Wheatstone full-bridge distribution on the cross beam is shown in Figure 5. When the turbulence signal acts on the cilium, the central mass will produce displacement due to the action of the force of inertia. The movement of the central mass will deform the cantilever beam, and the upper surface of both ends of the cantilever beam will be subjected to opposite stresses (Figure 6). That is, when two varistors on the cantilever beam are subjected to tensile stress, one is subjected to compressive stress, and the tensile stress and compressive stress change the resistance value of the varistor in the opposite direction; thus, one resistance increases and another resistance is reduced. This change will cause the original balance of the Wheatstone bridge to lose balance, and the change of stress on the cantilever beam will be converted into the change of the output voltage of the Wheatstone bridge through the varistor.

Sensitivity and resonant frequency are two important performance indexes to judge whether a probe is good or bad. When optimizing the design of the probe, it is necessary to improve the sensitivity performance of the probe on the premise of ensuring the detection bandwidth of the probe. The sensitivity of the MEMS cilium shear probe is proportional to its voltage output.

The mechanical analysis model of the microstructure is shown in Figure 6. It can be seen from Equation (1) that the probe output is proportional to the resistance stress on the beam, and the stress ${\sigma }_l$ is calculated as follows [21,22]:

$${\sigma }_l=\pm \frac{L^2+3aL-3\left(L+2a\right)x}{\frac{2}{3}b{t}^2\left(L^2+3aL+3a^2\right)}M\pm \frac{F}{bt}=\left[\pm \frac{L^2+3aL-3\left(L+2a\right)x}{\frac{2}{3}b{t}^2\left(L^2+3aL+3a^2\right)}{h}_0\pm \frac{1}{bt}\right]PS$$

(1)

where L is the length of the cantilever beam, a is the half length of the central mass block, b is the width of the cantilever beam, t is the thickness of the cantilever beam, x is the length of each point from the center of the beam, h₀ is the height of the cilium center of gravity, S is the receiving area of the turbulence signal, and P is the turbulent fluctuating pressure.

The resonant frequency of the probe microstructure affects the working frequency band of the shear probe. The resonant frequency of the probe is positively correlated with stiffness and is negatively correlated with mass. Through these analyses, we can find that the sensitivity and resonant frequency of the probe cannot be improved at the same time, which requires finding a suitable parameter to ensure the performance of the probe. At the same time, in order not to affect the deflection of the centroid, the bottom diameter of the cilium should not exceed the edge length of the mass block. Here, we select the photosensitive resin as the material of the cilium, which has a high Young’s modulus, and its density is very close to seawater, which can ensure that the probe can obtain a higher natural frequency and better detect turbulent vibration signals at the same time. In order to further improve the sensitivity of the probe, this study designed a bullet-like cilium structure, which can increase the receiving area of the probe for turbulence signals and improve the center of gravity of the cilium. In order to meet the requirements of both resonant frequency and sensitivity, it is necessary to establish a mathematical model to study the influence of probe-sensitive structural parameters on natural frequency and sensitivity through parametric sweeping.

2.2. Optimization of Structure Dimension

According to the process design requirements and design simulation, the size of the design beam is, for length, L = 1000 um, width W = 120 um, and thickness t = 40 um, and the side length of the mass block a = 300 um. Based on these dimensions, we performed a numerical simulation analysis of the size of cilium. COMSOL is a multiphysics simulation software based on advanced numerical calculation methods. It allows for parametric scans of cilium height and radius to determine the optimal size of the cilium by obtaining the cross-beam stress distribution and the relationship between the probe resonant frequency and height and radius, respectively. We used the control variable method: when scanning the cilium radius, cilium height was fixed to 5 mm and the radius of the scanning range is 130 um ≤ r ≤ 210 um. Similarly, when performing a parametric scan of cilium height, the cilium radius was set to 150 um, and the height scanning range was 2 mm ≤ h ≤ 10 mm. The simulation results are shown in Figure 7a,b, from which we can conclude that the stress on the cross beam is positively correlated with the cilium radius and height, and the larger the r and h, the greater the stress on the beam; the probe resonant frequency is negatively correlated with the cilium height and positively correlated with the radius. Greater height leads to greater eigenfrequency. Bigger radius leads to smaller eigenfrequency. Based on the scan conclusion, the values of cilium radius and height are determined: r = 175 mm, h = 6 mm. On this basis, the height of the bullet and the radius of the bottom are parametrically scanned. When scanning the height of the bullet, the fixed bottom radius is 750 um, and the height scanning range is 1000 um ≤ e ≤ 1800 um. In addition, when scanning the bottom radius of the bullet, the fixed height is 1500 um, and the bottom radius scanning range is 350 um ≤ d ≤ 1250 um. The simulation results are shown in Figure 7c,d, from which we can conclude that the stresses on the cross beam are negatively correlated with the height of the bullet and positively correlated with the bottom radius; both the probe resonant frequency and the bullet height are negatively correlated with the bottom radius. Select the appropriate parameters: e = 1500 um, d = 750 um. At this point, all the size parameters of the probe are determined.

Turbulence acts on the sensitive head of the shear probe, where shear force is created. Statistical analysis of turbulence can be performed via the signal caused by shear force. Hence, detecting resolution strongly depends on the diameter of the sensitive head. The sensitive head of the MEMS shear probe is 750 μm, much smaller than that of the commercial PNS probe (minimum 3 mm). Hence, the MEMS shear probe could realize a higher spatial resolution.

According to the determined parameter results, the three-dimensional model of the probe is established with COMSOL; the material properties are added according to the actual situation; the physics is set to select solid mechanics, and the boundary conditions are set; and the regional grid is divided according to the physics and finally placed in the air field and the water for steady-state and resonant frequency studies. Figure 8 shows the stress distribution curve on the X axis of the cross beam, and compared with the ordinary columnar cilium, it can be seen that the stress distribution on the beam of the bullet-headed probe is larger than that of the ordinary columnar cilium. To make it easier to reflect the stress changes, the varistor is placed on the central axis of the beam, near the roots at both ends of the beam, and the stress at this location reaches 2.53 × 10⁵ N/m², as shown in Figure 9. In addition, we performed a wet-model simulation for the probe eigenfrequency (Figure 10), and the first-order eigenfrequency of the probe is 490.9 Hz, which meets the needs of turbulence detection.

2.3. Micro-Manufacturing

The key sensitive unit of the probe is the crossbeam, which is manufactured by the MEMS process. The fabrication process is shown in Figure 11, mainly composed of four steps: (a) implant doping ion twice on SOI to form the piezoresistors; (b) pattern metal and anneal to form ohm contact and metallic connection lines; (c) dry etch device layer silicon to form crossbeam; (d) release crossbeam by etching substrate silicon thoroughly.

The first step of the fabrication process is photolithography. Before this process, the wafer is cleaned by a standardized cleaning process to eliminate organics and is preprocessed by hexamethyl disilazane (HMDS) to increase the adhesion of the silicon wafer and photoresist. An approximately 7 μm thick AZ 4620 photoresist (AZ Electronic Materials USA Corp, Branchburg, NJ, USA, a professional photoresist manufacturer mainly supplying AZ series photoresist) layer is spun on the silicon wafer, and then the piezoresistive zone mask is formed. Then, the piezoresistive zone is ion implanted (boron ion) to form a varistor.

The next step is sputtering the metal films. At the beginning, Cr and Au are sputtered on the SiO2 layer with about 20 and 100 nm layer thicknesses, successively. Then, a 2 μm thick AZ P5214 photoresist (AZ Electronic Materials USA Corp.) is spun and patterned onto the unit, which is formed after the cilium is pasted onto the central disk.

The third step is the front RIE etching. First, the AZ6130 photoresist is lithographed to form a cross beam-disc mask pattern, and then, the SiO2 buried oxygen layer is etched by using the RIE etching machine to form a cross-beam disc structure, and then the photoresist is removed.

The final step is the deep silicon etching on the back. First, the cross-beam pattern is lithographed with AZ6130 photoresist on the back of SOI, and then, the silicon is etched with a deep silicon etching machine, etching the silicon to the buried oxygen layer, releasing the cross-beam disc structure, and finally removing the photoresist.

2.4. Probe Packaging

The purpose of this paper was to propose a new type of shear probe based on a two-dimensional MEMS vector hydrophone. The MEMS vector hydrophone was researched and developed [23]. Its product, hydrophone, had high sensitivity and directivity, and a lot of work on how to improve the performance of hydrophone research has been conducted. The steel tube shell structure of the original hydrophone was changed, and the shell was prepared by 3D printing technology to reduce the weight and costs. The overall structure of the shear probe is shown in Figure 12a, and the sensitive unit of the shear probe is shown in Figure 12b.

We bonded the prepared probe chip to PCB with conductive silver paste and gold wire. After the lead bonding, under the optical experimental platform, we spliced the cilium to the disc in the center of the cross-beam structure with ultraviolet glue. Then, a 6 um parylene C film was deposited onto the probe by chemical vapor deposition technology to achieve waterproof insulation and corrosion protection [24]. In terms of probe housing packaging, this study proposed a streamlined package housing, which can effectively avoid the influence of eddy currents in water and enhance the diversion effect.

3. Results

3.1. Sensitivity Calibration Experiment

The sensitivity of the shear probe was calculated through a calibration method [25]. The PNS shear probe and bullet-headed shear probe were placed in the same position on the turbulence test platform (Figure 13a), and output voltage signals of the probes were measured under different flow shocks. The experimental results are shown in the figure, when other conditions are consistent. Sensitivity of MEMS shear probe can be calculated as follows:

$$\frac{S_P}{S_M}=\frac{V_{P\text{-}out}}{V_{M\text{-}out}}$$

(2)

where V_P-out and V_M-Sout are the root mean square value of the output voltage of the PNS probe and the MEMS shear probe, respectively, and S_P and S_M are the sensitivities, respectively.

The sensitivity of PNS is known to be 5.16 × 10⁻⁴ V·m·s²/kg. According to Equation (2), the sensitivity of the bullet-headed probe is calculated. The measured data curves are shown in Figure 13b–d, and the results and comparison of the different probes’ sensitivities are shown in

Table 1. Sensitivity of different probes.

Different Probe	Probe Sensitivity (V·m·s2/kg)
PNS [15]	5.16 × 10−4 V·m·s2/kg
Columnar cilium probe [19]	2.68 × 10−2 V·m·s2/kg
Lollipop-shaped sensor [20]	2.73 × 10−2 V·m·s2/kg
Bullet-headed probe	4.84 × 10−2 V·m·s2/kg

.

The sensitivity of the bullet-headed shear probe was 93.8 times that of PNS probe and 1.8 times that of the columnar cilium shear probe. There are two main reasons. The first one is the increased receiving area of the turbulence signal for the bullet cilium compared to the column cilium. The other one is the increased center of gravity of the bullet cilium. The bullet-headed shear probe shows advantages in observing turbulence signals.

3.2. Vector Test Experiment

Vector detection performance of MEMS shear probes is one of the main advantages over other ocean shear probes. At present, in ocean turbulence observation, one-dimensional shear probes are mainly used, and these probes measure in a single direction, which makes it difficult to observe the spatial characteristics of ocean turbulence. The MEMS shear probe in this study can naturally decompose the turbulence signal into two directions, X and Y, due to the cross-beam structure of its core sensitive element. To verify that the shear probe has two-dimensional vector performance, it is put into the turbulence test platform for vector performance testing. Before testing, the Reynolds number and turbulent flow direction in the turbulence test platform were simulated with COMSOL simulation software. The turbulence model type was chosen as the RANS Reynolds mean equation, and the turbulence model was chosen as the k-ε model [26]. The boundary conditions were set, and then, the steady-state study was performed. The simulation results are shown in Figure 14, and the parameters are shown in

Table 2. Description of flow field parameters.

ρ (kg/m3)	v (m/s)	μ (pa · s)	d (m)
1000	0.081	0.001	0.4

. The Reynolds number of the flow field surface was calculated to be 32,400, which indicates the presence of micro-scale turbulence within the flow field. The simulation results show that the flow velocity is 0.09 m/s in the X-direction and 0.005 m/s in the Y-direction in the position shown in the figure. There is a significant variability in the flow trend within the flow field, which demonstrates that the flow field of the test platform can be used for vectorial testing of the probe.

During the test, the probe was placed at the position in the simulation diagram so that the flow direction of the water was parallel to the X-way of the probe and perpendicular to the Y-way, and the collected data were calculated as the valid value to obtain Figure 15a. The figure shows that the probe X-way output is significantly larger than the Y-way output. To verify that this result was not determined by the structure of the probe itself, the probe was rotated 90° so that the water flow direction was parallel to the Y-way and perpendicular to the X-way. The test results are shown in Figure 15b. The figure shows that the Y-way output is larger than the X-way output, verifying the vector performance of the probe. The scatter plots of the acquired signals in Figure 15c,d visualize the vector performance of the bullet-headed MEMS shear probe.

4. Discussion

This paper presents a high-sensitivity bullet-headed MEMS shear probe with improved structural dimensions. The bullet-headed cilium increases the receiving area of the cilium and raises the center of gravity of the cilium, which makes it easier for the probe to receive the weak turbulence signal. For the ocean turbulence surroundings, the wet-mode simulation of the probe was carried out to make the frequency response bandwidth of the probe meet the needs of ocean turbulence detection. The sensitivity of the probe was increased to 4.84 × 10⁻² V·m·s²/kg, which is 1.8 times as high as that of the columnar cilium shear probe and 93.8 times that of the PNS shear probe. The vector test also validates the capacity of the probe in the detection of two-dimensional turbulence signals, which is of great significance to ocean turbulence observation. The probe could not bear any collision directly acted on the cilium, which may cause fracture of the silicon beams. During the use and preservation of the probe, we needed to be careful to avoid damage of the probe by external collision, especially the cilium. An additional protective cap can be placed on the head of the probe. The robustness of the MEMS probe could be enhanced in future research. In conclusion, the bullet-headed MEMS shear probe proposed in this paper has the advantages of high sensitivity, high-spatial resolution, and two-dimensional vector observation, which is suitable for autonomous profiling observation of marine turbulence.